Clean up validate_llama_ref_model example and requirements for examples

requirements.txt

@@ -3,6 +3,11 @@ gguf==0.6.0
 numpy==1.26.3
 onnx==1.15.0
 
+# Model deps.
+huggingface-hub==0.22.2
+transformers==4.40.0
+sentencepiece==0.2.0
+
 # It is expected that you have installed a PyTorch version/variant specific
 # to your needs, so we only include a minimum version spec.
 # TODO: Use a versioned release once 2.3.0 drops.

sharktank/sharktank/examples/validate_llama_ref_model.py

@@ -14,8 +14,16 @@
 
 
 def main(args: list[str]):
+    from ..utils import cli
+
     torch.no_grad().__enter__()
-    config = Dataset.load(args[0])
+
+    parser = cli.create_parser()
+    cli.add_gguf_dataset_options(parser)
+    args = cli.parse(parser)
+
+    data_files = cli.get_gguf_data_files(args)
+    config = Dataset.load(data_files["gguf"])
     hp = configs.LlamaHParams.from_gguf_props(config.properties)
     model = DirectCacheLlamaModelV1(config.root_theta, hp)
 

sharktank/sharktank/models/llama/llama_ref.py

@@ -13,6 +13,7 @@
 import torch.nn.functional as F
 
 from ...layers import *
+from ...types import Theta
 
 __all__ = [
     "DirectCacheLlamaModelV1",